From: Mohan Kumar M <mohan@in.ibm.com>
To: Randy Dunlap <randy.dunlap@oracle.com>
Cc: linuxppc-dev@ozlabs.org, fastboot@lists.osdl.org,
anton@samba.org, paulus@samba.org
Subject: Re: [Fastboot] [PATCH] Avoid hypervisor statistics calculation in real mode
Date: Wed, 21 Mar 2007 11:21:32 +0530 [thread overview]
Message-ID: <20070321055132.GA4200@in.ibm.com> (raw)
In-Reply-To: <20070320153751.462da242.randy.dunlap@oracle.com>
Thanks Randy.
I updated the patch as per your suggestions.
kexec invokes plpar_hcall hypervisor call in real mode. plpar_hcall
refers per cpu variable for accounting hypervisor statistics. These
variables may not be present in the RMO region. So it results in
0x300 exception.
The following patch fixes this problem by using raw hypervisor call
which does not update the hypervisor call statistics. Thanks to Anton
for suggesting this idea.
Cc: Anton Blanchard <anton@samba.org>
Signed-off-by: Mohan Kumar M <mohan@in.ibm.com>
---
arch/powerpc/platforms/pseries/hvCall.S | 34 ++++++++++++++++++++++++
arch/powerpc/platforms/pseries/lpar.c | 2 -
arch/powerpc/platforms/pseries/plpar_wrappers.h | 16 +++++++++++
include/asm-powerpc/hvcall.h | 14 +++++++++
4 files changed, 65 insertions(+), 1 deletion(-)
Index: linux-2.6.21-rc4/arch/powerpc/platforms/pseries/hvCall.S
===================================================================
--- linux-2.6.21-rc4.orig/arch/powerpc/platforms/pseries/hvCall.S
+++ linux-2.6.21-rc4/arch/powerpc/platforms/pseries/hvCall.S
@@ -123,6 +123,40 @@ _GLOBAL(plpar_hcall)
blr /* return r3 = status */
+/*
+ * plpar_hcall_raw can be called in real mode. kexec/kdump need some
+ * hypervisor calls to be executed in real mode. So plpar_hcall_raw
+ * does not access the per cpu hypervisor call statistics variables,
+ * since these variables may not be present in the RMO region.
+ */
+_GLOBAL(plpar_hcall_raw)
+ HMT_MEDIUM
+
+ mfcr r0
+ stw r0,8(r1)
+
+ std r4,STK_PARM(r4)(r1) /* Save ret buffer */
+
+ mr r4,r5
+ mr r5,r6
+ mr r6,r7
+ mr r7,r8
+ mr r8,r9
+ mr r9,r10
+
+ HVSC /* invoke the hypervisor */
+
+ ld r12,STK_PARM(r4)(r1)
+ std r4, 0(r12)
+ std r5, 8(r12)
+ std r6, 16(r12)
+ std r7, 24(r12)
+
+ lwz r0,8(r1)
+ mtcrf 0xff,r0
+
+ blr /* return r3 = status */
+
_GLOBAL(plpar_hcall9)
HMT_MEDIUM
Index: linux-2.6.21-rc4/arch/powerpc/platforms/pseries/lpar.c
===================================================================
--- linux-2.6.21-rc4.orig/arch/powerpc/platforms/pseries/lpar.c
+++ linux-2.6.21-rc4/arch/powerpc/platforms/pseries/lpar.c
@@ -378,7 +378,7 @@ static void pSeries_lpar_hptab_clear(voi
/* TODO: Use bulk call */
for (i = 0; i < hpte_count; i++)
- plpar_pte_remove(0, i, 0, &dummy1, &dummy2);
+ plpar_pte_remove_raw(0, i, 0, &dummy1, &dummy2);
}
/*
Index: linux-2.6.21-rc4/arch/powerpc/platforms/pseries/plpar_wrappers.h
===================================================================
--- linux-2.6.21-rc4.orig/arch/powerpc/platforms/pseries/plpar_wrappers.h
+++ linux-2.6.21-rc4/arch/powerpc/platforms/pseries/plpar_wrappers.h
@@ -78,6 +78,22 @@ static inline long plpar_pte_remove(unsi
return rc;
}
+/* plpar_pte_remove_raw can be called in real mode. It calls plpar_hcall_raw */
+static inline long plpar_pte_remove_raw(unsigned long flags, unsigned long ptex,
+ unsigned long avpn, unsigned long *old_pteh_ret,
+ unsigned long *old_ptel_ret)
+{
+ long rc;
+ unsigned long retbuf[PLPAR_HCALL_BUFSIZE];
+
+ rc = plpar_hcall_raw(H_REMOVE, retbuf, flags, ptex, avpn);
+
+ *old_pteh_ret = retbuf[0];
+ *old_ptel_ret = retbuf[1];
+
+ return rc;
+}
+
static inline long plpar_pte_read(unsigned long flags, unsigned long ptex,
unsigned long *old_pteh_ret, unsigned long *old_ptel_ret)
{
Index: linux-2.6.21-rc4/include/asm-powerpc/hvcall.h
===================================================================
--- linux-2.6.21-rc4.orig/include/asm-powerpc/hvcall.h
+++ linux-2.6.21-rc4/include/asm-powerpc/hvcall.h
@@ -237,6 +237,20 @@ long plpar_hcall_norets(unsigned long op
long plpar_hcall(unsigned long opcode, unsigned long *retbuf, ...);
/**
+ * plpar_hcall_raw: - Make a hypervisor call without calculating hcall stats
+ * @opcode: The hypervisor call to make.
+ * @retbuf: Buffer to store up to 4 return arguments in.
+ *
+ * This call supports up to 6 arguments and 4 return arguments. Use
+ * PLPAR_HCALL_BUFSIZE to size the return argument buffer.
+ *
+ * Used when phyp interface needs to be called in real mode. Similar to
+ * plpar_hcall, but plpar_hcall_raw works in real mode and does not
+ * calculate hypervisor call statistics.
+ */
+long plpar_hcall_raw(unsigned long opcode, unsigned long *retbuf, ...);
+
+/**
* plpar_hcall9: - Make a pseries hypervisor call with up to 9 return arguments
* @opcode: The hypervisor call to make.
* @retbuf: Buffer to store up to 9 return arguments in.
next prev parent reply other threads:[~2007-03-21 5:52 UTC|newest]
Thread overview: 5+ messages / expand[flat|nested] mbox.gz Atom feed top
2007-03-20 13:35 [PATCH] Avoid hypervisor statistics calculation in real mode Mohan Kumar M
2007-03-20 22:37 ` [Fastboot] " Randy Dunlap
2007-03-21 5:51 ` Mohan Kumar M [this message]
2007-03-21 15:30 ` Randy Dunlap
2007-03-23 6:18 ` Horms
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20070321055132.GA4200@in.ibm.com \
--to=mohan@in.ibm.com \
--cc=anton@samba.org \
--cc=fastboot@lists.osdl.org \
--cc=linuxppc-dev@ozlabs.org \
--cc=paulus@samba.org \
--cc=randy.dunlap@oracle.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).